PhD candidate announces MIT thesis defense on physics-constrained generative models
The candidate argues that generating designs quickly isn't enough: a design becomes a solution only if it can be realized in the real world and satisfies the design problem's functional requirements.
TLDR
A PhD candidate announced a September 8, 2026, defense of “Physics-Constrained Generative Models for Computational Design” at MIT CSAIL. The September 7 announcement listed Kaiming He, Bill Freeman and Wojciech Matusik as committee members and shared a Zoom link. The candidate's central point: generative models can propose designs at high throughput, but those designs must be physically realizable and meet functional requirements to count as solutions.
Combined views
—
2 Sources, first seen ago
